We analyze the feasibility of so-called radial surface currents as derived from Sentinel-1 Wave mode Doppler shift observations to support the geodetic estimation of the mean dynamic topography ...(MDT). Broadly defined as subtracting the geoid from the mean sea surface, this separation problem suffers from data inhomogeneities and usually requires strong prior assumptions about the MDT’s smoothness. Recent advancements in calibration and current retrieval methodology for Sentinel-1 data give reason to assess potential gains from this observation type, which can be linked to the gradient of the MDT under geostrophic approximation. Due to current lack of long time-series, we synthesize 10 years of observations from daily surface currents grids and include these in a parametric least-squares adjustment of the MDT. Our results utilizing the synthetic data in place of smoothness constraints are promising, showing 2 cm RMS agreement with a state-of-the-art MDT model.
Simple stacking of InSAR Digital Elevation Models (DEMs) has shown potential to increase DEM quality. In order to obtain a straight-forward stacking procedure that reduces the uncertainty of InSAR ...DEMs, in this contribution we evaluate in detail the impact of different stacking routines on the accuracy of stacked DEMs. For that end, we performed systematic tests in a region of Córdoba, Argentina. First, we produced a set of 54 Sentinel-1 and 10 SAOCOM-1 InSAR DEMs, which were evaluated with respect to a reference photogrammetric DEM. We then tested different stacking workflows, obtaining stacked DEMs with a higher average accuracy than that of the single-pair DEMs. This suggests that the uncertainty in the quality of the InSAR DEMs can be overcome by simple stacking techniques. Further post-processing of the stacked DEMs involved planimetric position correction by co-registration with the reference DEM, altimetric correction by linear regression adjustment to the national altimetric network, and multidirectional filtering to correct for speckling and outliers. The evaluation of the final stacked DEMs with respect to the reference shows that the SAOCOM-1 stacked DEM has a mean biased error of 0.39 m with a standard deviation of 3.77 m, whereas the Sentinel-1 stacked DEM has a mean biased error of 2.25 m with a standard deviation of 7.61 m. Both DEMs offered a smaller pixel size (15 m) than the available Argentine digital elevation model MDE-Ar (30 m), but with a lower accuracy. In turn, the combination of the SAOCOM-1 and Sentinel-1 stacked DEMs resulted in a 50 % and 33 % reduction in the mean biased error and the standard deviation, respectively, with respect to the SAOCOM-1 stacked DEM; an accuracy close to that of MDE-Ar, with a smaller pixel size. Although further improvements could be accomplished by exploring more sophisticated stacking and data-fusion techniques, these results constitute a significant step towards the systematization of a methodology to obtain reliable DEMs from SAR data.
•Simple stacking constitutes a reliable solution to overcome the quality uncertainty of InSAR DEMs.•SAOCOM-1 data proved to be useful for production of medium resolution, high quality DEMs.•The weighted combination of SAOCOM-1 and Sentinel-1 stacks improved DEM accuracy in terms of bias and standard deviation.
•Wetland complexities were mapped using a multi-level classification scheme.•Both wetland vegetation types and surface water dynamics were classified.•Combining Sentinel-1 and -2 provided the most ...accurate results.•High-vegetated wetlands could not be mapped accurately using Sentinel-1 and -2.
Wetlands have been determined as one of the most valuable ecosystems on Earth and are currently being lost at alarming rates. Large-scale monitoring of wetlands is of high importance, but also challenging. The Sentinel-1 and -2 satellite missions for the first time provide radar and optical data at high spatial and temporal detail, and with this a unique opportunity for more accurate wetland mapping from space arises. Recent studies already used Sentinel-1 and -2 data to map specific wetland types or characteristics, but for comprehensive wetland characterisations the potential of the data has not been researched yet. The aim of our research was to study the use of the high-resolution and temporally dense Sentinel-1 and -2 data for wetland mapping in multiple levels of characterisation. The use of the data was assessed by applying Random Forests for multiple classification levels including general wetland delineation, wetland vegetation types and surface water dynamics. The results for the St. Lucia wetlands in South Africa showed that combining Sentinel-1 and -2 led to significantly higher classification accuracies than for using the systems separately. Accuracies were relatively poor for classifications in high-vegetated wetlands, as subcanopy flooding could not be detected with Sentinel-1’s C-band sensors operating in VV/VH mode. When excluding high-vegetated areas, overall accuracies were reached of 88.5% for general wetland delineation, 90.7% for mapping wetland vegetation types and 87.1% for mapping surface water dynamics. Sentinel-2 was particularly of value for general wetland delineation, while Sentinel-1 showed more value for mapping wetland vegetation types. Overlaid maps of all classification levels obtained overall accuracies of 69.1% and 76.4% for classifying ten and seven wetland classes respectively.
Continuous monitoring of reservoirs and dams is essential for efficient water management. Synthetic Aperture Radar (SAR) imagery offers the potential for continuous monitoring of surface water ...through all-weather ground observation. The objective of this study is to enhance the accuracy of water body detection and water quantity estimation by applying 64 combinations of speckle filtering and object detection techniques to Sentinel-1 imagery. For speckle filtering, the Median, Gaussian, Lee, and Frost techniques were used with various window sizes (3, 5, 7, and 9). For water body detection, the Otsu, Kittler-Illingworth (KI), Chan-Vese (CV), and K-means methods were employed. The study area included three reservoirs and two dams in Korea, encompassing a variety of water surface sizes and types of land cover. To validate the accuracy of each water body detection combination, manual delineation-based water mask images from Sentinel-2 were employed. Furthermore, a regression equation (y=axb) between water surface area and storage was used to estimate water storage based on SAR imagery, followed by time-series validation using in-situ data. The research results indicate that the optimal detection technique varies significantly depending on the type of surrounding land cover and the size of the water body. The highest performance was observed for the CV technique combination for waterfront pixels, and for the KI technique combination for other land cover pixels. In speckle filtering techniques using a large window size, the false detection rate caused by vegetation and buildings was low; however, the boundaries of water bodies were blurred. Consequently, using smaller window sizes in SAR imagery and leveraging optimal water body detection combinations specific to land cover types, along with post-processing using masking data, would enhance the performance of water surface area and storage estimation.
Display omitted
•Reservoirs and dams are monitoring based on SAR images of the Korean peninsula.•64 combinations of speckle filtering and object detection techniques are compared.•Optimal water body detection combination varies by land cover type.•Small window size speckle filtering with masking yields the best results.
•A machine-learning based high-resolution mapping workflow was proposed.•Canopy height product from ICESat-2 satellite was validated by airborne LiDAR data.•Deep-learning and random forest models ...were used to upscale ICESat-2 canopy height.•Sentinel-2 and Landsat-8 data were compared in the prediction of forest height.•Sentinel-1 & -1 satellites showed high capacity on the prediction of forest height.
Forest canopy height is an important indicator of forest carbon storage, productivity, and biodiversity. The present study showed the first attempt to develop a machine-learning workflow to map the spatial pattern of the forest canopy height in a mountainous region in the northeast China by coupling the recently available canopy height (Hcanopy) footprint product from ICESat-2 with the Sentinel-1 and Sentinel-2 satellite data. The ICESat-2 Hcanopy was initially validated by the high-resolution canopy height from airborne LiDAR data at different spatial scales. Performance comparisons were conducted between two machine-learning models – deep learning (DL) model and random forest (RF) model, and between the Sentinel and Landsat-8 satellites. Results showed that the ICESat-2 Hcanopy showed the highest correlation with the airborne LiDAR canopy height at a spatial scale of 250 m with a Pearson’s correlation coefficient (R) of 0.82 and a mean bias of -1.46 m, providing important evidence on the reliability of the ICESat-2 vegetation height product from the case in China’s forest. Both DL and RF models obtained satisfactory accuracy on the upscaling of ICESat-2 Hcanopy assisted by Sentinel satellite co-variables with an R-value between the observed and predicted Hcanopy equalling 0.78 and 0.68, respectively. Compared to Sentinel satellites, Landsat-8 showed relatively weaker performance in Hcanopy prediction, suggesting that the addition of the backscattering coefficients from Sentinel-1 and the red-edge related variables from Sentinel-2 could positively contribute to the prediction of forest canopy height. To our knowledge, few studies have demonstrated large-scale vegetation height mapping in a resolution ≤ 250 m based on the newly available satellites (ICESat-2, Sentinel-1 and Sentinel-2) and DL regression model, particularly in the forest areas in China. Thus, the present work provided a timely and important supplementary to the applications of these new earth observation tools.
The 1000-km-long Haiyuan fault system on the northeastern edge of the Tibetan Plateau contributes to accommodating the deformation in response to the India/Asia collision. In spite of its importance, ...the kinematics of the fault including the geometry and along-strike slip rate have not been completely defined. In this study, we use synthetic aperture radar data acquired between 2014 and 2021 by Sentinel-1 satellites to investigate the present-day strain accumulation on the Haiyuan fault system. We produce a high-resolution velocity map for the ∼300,000 km2 Haiyuan region using the Small BAseline Subset method. Our new velocity fields reveal deformation patterns dominated by the eastward motion of Tibet relative to Alaxan and localised strain accumulation along the Haiyuan, Gulang and Xiangshan-Tianjingshan faults. The western ∼300-km-long section of the Haiyuan fault, which was previously unmapped, seems to follow Tuolaishan and terminate at Halahu. We compute the along-strike slip rate using a Bayesian Markov Chain Monte Carlo inversion approach, and find that the overall strike-slip rate along the Haiyuan fault system gradually increases from the western end (1.8±0.3 mm/yr close to Halahu) to the east (6.4±0.5 mm/yr before entering Liupanshan), and further east, it decreases from 6.4±0.5 mm/yr to 1.3±0.7 mm/yr. The Haiyuan fault absorbs most of the left-lateral strike-slip motion with a rate of ∼4.2±0.4 mm/yr, and the Gulang and Xiangshan-Tianjingshan faults take up a fraction of 2.2±0.6 mm/yr. We re-map the previously identified shallow creeping zone on the Laohushan segment for a length of 45 km, slightly larger than the previous estimate of 35 km. The average shallow creep rate, 3 mm/yr between 2014–2021, is consistent with the rate before 2007 (2–3 mm/yr), implying that the shallow creep is a steady behaviour.
•We produce a high-resolution InSAR velocity map for the ∼300,000 km2 Haiyuan region.•The western section of the Haiyuan fault follows Tuolaishan and terminates at Halahu.•Strike-slip rate gradually increases from the western end (∼1.8 mm/yr) to the east.•We re-map the shallow creeping zone on the Laohushan segment for a length of 45 km.
Marine oil spills pose significant ecological and economic threats worldwide, requiring effective decision-making tools. In this study, the optimal parameters, and configurations for Deep Learning ...models in oil spill classification and segmentation using Sentinel-1 SAR imagery were identified. First, a new Sentinel-1 image dataset was created. Ninety CNN configurations were explored for classification by varying the number of convolutional layers, filters, hidden layers, and neurons in each layer. For segmentation tasks, MLP and U-Net models were evaluated with variations in convolutional layers, filters, and incorporation of IoU and Focal Loss. The results indicated that a CNN model with six layers, 32 filters, and two hidden layers achieved 99 % classification accuracy. For segmentation, the U-Net model with more layers and filters using Focal Loss achieved 99 % accuracy and 96 % IoU. Therefore, a CNN and U-Net framework was proposed that achieves an overall accuracy of 95 % and an IoU of 90 %.
Display omitted
•Optimizing CNN configurations achieves 99 % accuracy in oil spill detection.•Pixel-based MLP models for oil spill segmentation generate false-positives.•U-Net architecture optimization resulted in an IoU of 96 % for segmentation.•Focal Loss and IoU improves oil spill segmentation certainty.•Classification and segmentation scheme reduces false positives in look-alikes scenarios.
Soil organic carbon (SOC) and soil total nitrogen (STN) are important indicators of soil health and play a key role in the global carbon and nitrogen cycles. High-resolution radar Sentinel-1 and ...multispectral Sentinel-2 images have the potential to investigate soil spatial distribution information over a large area, although Sentinel-1 and Sentinel-2 data have rarely been combined to map either SOC or STN content. In this study, we applied machine learning techniques to map both SOC and STN content in the southern part of Central Europe using digital elevation model (DEM) derivatives, multi-temporal Sentinel-1 and Sentinel-2 data, and evaluated the potential of different remote sensing sensors (Sentinel-1 and Sentinel-2) to predict SOC and STN content. Four machine-learners including random forest (RF), boosted regression trees (BRT), support vector machine (SVM) and Bagged CART were used to construct predictive models of SOC and STN contents based on 179 soil samples and different combinations of environmental covariates. The performance of these models was evaluated based on a 10-fold cross-validation method by three statistical indicators. Overall, the BRT model performed better than RF, SVM and Bagged CART, and these models yielded similar spatial distribution patterns of SOC and STN. Our results showed that multi-source sensor methods provided more accurate predictions of SOC and STN contents than individual sensors. The application of radar Sentinel-1 and multispectral Sentinel-2 images proved useful for predicting SOC and STN. A combination of Sentinel-1/2-derived predictors and DEM derivatives yielded the highest prediction accuracy. The prediction accuracy changed with and without the Sentinel-1/2-derived predictors, with the R2 for estimating both SOC and STN content using the BRT model increasing by 12.8% and 18.8%, respectively. Topographic variables were the main explanatory variables for SOC and STN predictions, where elevation was assigned as the variable with the most importance by the models. The results of this study illustrate the potential of free high-resolution radar Sentinel-1 and multispectral Sentinel-2 data as input when developing SOC and STN prediction models.
Display omitted
•Multi-source sensor methods achieved more accurate SOC and STN predictions than single sensors.•The potential of Sentinel-1 and 2 data in predicting SOC and STN was explored.•Boosted regression trees model performed best in predicting SOC and STN.
•Wetland classification is challenging, but essential.•Dense optical and SAR data help better identification of wetland vegetation types.•An object-based stacked generalization method for wetland ...mapping is proposed.•The comparison between pixel-based and object-based method is performed in wetland mapping.•The stability of the proposed method is evaluated.
Wetland ecosystems have experienced dramatic challenges in the past few decades due to natural and human factors. Wetland maps are essential for the conservation and management of terrestrial ecosystems. This study is to obtain an accurate wetland map using an object-based stacked generalization (Stacking) method on the basis of multi-temporal Sentinel-1 and Sentinel-2 data. Firstly, the Robust Adaptive Spatial Temporal Fusion Model (RASTFM) is used to get time series Sentinel-2 NDVI, from which the vegetation phenology variables are derived by the threshold method. Subsequently, both vertical transmit-vertical receive (VV) and vertical transmit-horizontal receive (VH) polarization backscatters (σ0 VV, σ0 VH) are obtained using the time series Sentinel-1 images. Speckle noise inherent in SAR data, resulting in over-segmentation or under-segmentation, can affect image segmentation and degrade the accuracies of wetland classification. Therefore, we segment Sentinel-2 multispectral images to delineate meaningful objects in this study. Then, in order to reduce data redundancy and computation time, we analyze the optimal feature combination using the Sentinel-2 multispectral images, Sentinel-2 NDVI time series, phenological variables and other vegetation index derived from Sentinel-2 multispectral images, as well as time series Sentinel-1 backscatters at the object level. Finally, the stacked generalization algorithm is utilized to extract the wetland information based on the optimal feature combination in the Dongting Lake wetland. The overall accuracy and Kappa coefficient of the object-based stacked generalization method are 92.46% and 0.92, which are 3.88% and 0.04 higher than that using the pixel-based method. Moreover, the object-based stacked generalization algorithm is superior to single classifiers in classifying vegetation of high heterogeneity areas.
PDF
